There is a storage cluster as one of technologies that improve reliability of storage devices. The storage cluster has a cluster structure constituted by a plurality of storage devices and is a technology that implements a failover in which, even if one of the storage devices is down due to a failure at a device level, business can be continued by another storage device.
The storage cluster can be set in units of volumes and automatically takes over Write/Read I/O of a storage device in which a failure has occurred to another storage device. Accordingly, this makes it possible for a user to continue the business without being aware of the occurrence of an abnormality and thus it is possible to reduce the recovery time and the workload imposed on a storage administrator.
A volume in the storage cluster can be implemented by allowing a host to indicate the volume as, for example, another path of the same volume. The “host” mentioned here is a device, such as a server or the like, that accesses the storage device, whereas the “path” mentioned here is a communication path between the host and the storage device. In the storage cluster, the path in operation is used as “active”, the path on the waiting side is used as “standby”, and only the active storage device can be accessed. If the storage device in operation is down, the storage cluster changes a path state, sets the path on the waiting side to active, and automatically takes over Write/Read I/O.
Furthermore, there is a conventional technology in which an off host Virtualizer creates meta data on an operating system of a first virtual storage device and can access a first host, thereby allowing a cross-platform access to the storage volumes.
Furthermore, there is a conventional technology in which, between data storage subsystems, by copying attribute information by using the same data transfer path as that used for a remote copy function of data, a host computer can use a data storage subsystem at the copy destination without handling details.
Patent Document 1: Japanese National Publication of International Patent Application No. 2007-516523
Patent Document 2: Japanese Laid-open Patent Publication No. 2004-13367
When an operation is performed with respect to a volume, a host converts a device name, of an operation target, that is commonly set for the storage devices to a logical unit number (LUN) and specifies the operation target by the LUN. The “LUN” mentioned here is an identification number for logically identifying a device for storage in the storage device.
Furthermore, in the plurality of storage devices constituting volumes in the storage cluster, the device names are made to match between the active and the standby storage devices; however, the LUNs do not need to be matched. The reason for this is that, in order to match LUNs of the active storage device and the standby storage device, the same configuration is needed in both the storage devices; therefore, the load imposed on an administrator is significantly increased. Accordingly, in the present state, LUNs are not made to match in most cases even if storage devices have association relationship of the active side and the standby side.
However, if a failover occurs when an operation of specifying a LUN of a volume in a storage cluster is performed, because the LUNs do not match between the volumes associated with the active side and the standby side, there is a problem in that a false volume is operated after the failover.